Verke Editorial

Research behind Verke

Two of Verke's coaches have been evaluated in a randomized controlled trial at Stockholm University, published in Internet Interventions (2026).

This page is a transparent look at the research behind Verke — the academic study the product was evaluated in, the peer-reviewed literature behind each coaching method, and the specific things we do and don't claim. Where the research is strong we say so; where it's developing we hedge; where results are published, we point you to the paper rather than restate them.

Two of our coaches were evaluated in a randomized controlled trial at Stockholm University, supervised by a leading internet-interventions researcher and published in a peer-reviewed journal; and the six coaching methods Verke uses each have their own independent peer-reviewed evidence base.

Two of Verke's coaches — Anna (psychodynamic) and Judith (cognitive-behavioral) — were evaluated in a three-arm randomized controlled trial at Stockholm University. One hundred and two adults with social anxiety disorder were randomized to AI-delivered psychodynamic coaching with Anna, AI-delivered cognitive-behavioral coaching with Judith, or a waitlist control group. The active intervention ran for four weeks, with the main symptom measures taken at the end of the intervention and again at a one-month follow-up.

The main endpoint is social anxiety symptom reduction measured with validated questionnaires. Secondary measures include depression symptoms, general anxiety, therapeutic alliance, and participant satisfaction.

Recruitment took place in 2025, and the results were published in the peer-reviewed journal Internet Interventions in 2026. The trial sits inside a broader research program at Stockholm University's Department of Psychology, supervised by Professor Per Carlbring. The project page is on the Stockholm University research catalogue: Stockholm University research catalogue, and the supervising researcher's site is carlbring.se.

The trial produced two student theses at Stockholm University's Department of Psychology, both supervised by Professor Carlbring and both publicly defended in 2025. They look at the same project from two different angles — one quantitative, one qualitative — and together cover the two questions most readers care about: how the trial was analysed, and what it was like to use the coaches.

Hassbrink (2025) — the quantitative thesis

Karla Hassbrink's master's thesis, Anna eller Judith: Artificiell intelligens vid social ångest, reports the trial's symptom-outcome analyses — between-arm differences at the end of the intervention and at the one-month follow-up. The thesis is a psykologexamensarbete (30 hp) defended at Stockholm University, autumn 2025, and is listed on Carlbring's public publications page (carlbring.se/uppsatser). It is the public source for the trial's primary-outcome reporting at thesis level.

Lutovsky (2025) — the qualitative thesis

Catharina Lutovsky's master's thesis, “Det är nästan som ett journummer fast det är AI” (“It's almost like a helpline, except it's AI”), is a qualitative reflexive-thematic-analysis study based on eleven semi-structured interviews with active-treatment participants from the same trial. Also a psykologexamensarbete (30 hp) at Stockholm University, supervised by Professor Carlbring, defended autumn 2025.

Lutovsky's analysis surfaces four broad themes from participants' experience of receiving AI-delivered therapy:

• The non-judgmental quality of the AI as a felt experience — and what that lets people bring up that they would otherwise not raise.
• Constant availability as both a strength (support at hard moments outside business hours) and a risk (over-reliance, blurred boundaries).
• How participants relate to mistakes — the specific ways an AI can misunderstand or forget context, and the implications for trust and shared responsibility.
• AI-supported insight and change — how participants used the conversations to understand themselves better, notice patterns and try new things.

We describe Lutovsky's themes at the conceptual level only. Direct participant quotes belong in her thesis, not in our marketing.

Professor Per Carlbring is Professor of Clinical Psychology at Stockholm University and a pioneer of internet-delivered psychotherapy research. He is Editor-in-Chief of the academic journal Internet Interventions, which he co-founded in 2014, and his research group has produced foundational randomized trials on internet-delivered CBT and psychodynamic therapy for anxiety disorders. With more than 300 peer-reviewed publications spanning iCBT, psychodynamic internet therapy, virtual-reality exposure, gambling-disorder treatment, and — increasingly since 2023 — AI-assisted psychotherapy, his lab is one of Europe's most productive in digital mental health. Carlbring supervised the Stockholm University study.

Carlbring's own peer-reviewed work has helped frame the legitimacy of AI in internet-delivered psychotherapy as a research direction — for example, his 2023 paper A new era in Internet interventions: The advent of Chat-GPT and AI-assisted therapist guidance (Carlbring et al., 2023, Internet Interventions) opened the conversation about how large-language-model coaches might fit alongside guided iCBT. We cite this as independent peer-reviewed framing of the broader research area Verke operates in, not as endorsement of Verke specifically.

Public references: Stockholm University profile · Google Scholar · publikationer (full publication list) · uppsatser (supervised theses, including Hassbrink 2025) · carlbring.se.

Each of the six coaching methods below has its own independent peer-reviewed evidence base, separate from the Stockholm University study. These are the methods our coaches are trained on. The method-specific pages go deeper into how each one works inside Verke; the summaries here are just the evidence anchor.

Cognitive Behavioral Therapy (CBT)

CBT is among the most researched psychological approaches for common mental-health concerns. Network meta-analyses identify individual CBT as one of the most effective psychological interventions for social anxiety disorder (Mayo-Wilson et al., 2014). Read more about CBT at Verke →

Psychodynamic Therapy (PDT)

PDT has been shown to be efficacious for social anxiety in head-to-head randomized trials against CBT, with both approaches showing durable benefit at long-term follow-up (Leichsenring et al., 2013). A 2023 umbrella review concluded PDT now meets contemporary criteria as an empirically-supported approach for depression, anxiety, and somatic symptom disorders (Leichsenring et al., 2023). Read more about PDT at Verke →

Acceptance and Commitment Therapy (ACT)

A meta-analysis of 39 randomized trials found ACT superior to waitlist, psychological placebo, and usual care across anxiety, depression, substance use, and chronic pain (A-Tjak et al., 2015). Read more about ACT at Verke →

Emotionally Focused Therapy (EFT)

EFT is an American Psychological Association-designated empirically supported approach for relationship distress, with meta-analyses reporting large benefits for couple distress across randomized trials (Wiebe & Johnson, 2016). Read more about EFT at Verke →

Compassion-Focused Therapy (CFT)

CFT has a growing evidence base, with meta-analyses reporting small-to-moderate effects on depression, anxiety, self-criticism, and self-compassion in clinical samples (Vidal & Soldevilla, 2023). The evidence base is smaller than for CBT or PDT and is still developing. Read more about CFT at Verke →

Nonviolent Communication (NVC)

NVC is a structured communication practice — observe, feel, need, request — rather than a clinical intervention, and the peer-reviewed evidence base is smaller and more methodologically mixed than the other methods on this list. Scoping reviews suggest positive effects on empathy, workplace conflict, and interpersonal communication (Museux et al., 2022). Read more about NVC at Verke →

For a consumer-friendly overview of how these modalities show up in AI coaching at Verke, see Types of AI therapy at Verke.

Verke's coaching methodology is informed by the clinical-psychology literature and refined against real user feedback at scale. The Stockholm University randomized trial was supervised by Professor Per Carlbring. Two student theses from the trial have been publicly defended at Stockholm University in 2025: the Hassbrink thesis (quantitative, symptom-outcome analyses, listed on carlbring.se/uppsatser) and the Lutovsky thesis (qualitative, post-treatment interviews and thematic analysis, defended autumn 2025 at Stockholm University's Department of Psychology). The senior research team's peer-reviewed journal article was published in Internet Interventions in 2026.

Verke's psychodynamic method draws on peer-reviewed work from Carlbring's research network at Stockholm University — including the 2017 internet-delivered affect-focused PDT trial by Johansson and colleagues (Johansson et al., 2017) and the 2024 guided-vs-unguided internet-delivered PDT trial by Lindegaard and colleagues (Lindegaard et al., 2024). Together, that body of work has established internet-delivered psychodynamic therapy as an evidence-supported approach for social anxiety. We cite this network as independent peer-reviewed support for the method Anna uses; its authors are not Verke team members or endorsers.

What we claim

• Verke delivers coaching using methods with substantial peer-reviewed evidence bases — CBT, PDT, ACT, EFT, CFT — and the structured communication practice of NVC.
• The Anna and Judith coaches in the Verke product are the same two AI profiles evaluated in the Stockholm University study.
• The Stockholm University randomized controlled trial is complete, and its results were published in the peer-reviewed journal Internet Interventions in 2026.
• Two complementary master's theses from the trial — Hassbrink (quantitative) and Lutovsky (qualitative) — have been publicly defended at Stockholm University in 2025.

What we don't claim

• Verke is not a replacement for professional mental-health care, psychiatric evaluation, or crisis support. It is a coaching product.
• Verke is an FDA-cleared medical device. It is not, and we make no medical-device claims.
• Specific clinical outcome numbers — symptom-reduction percentages, recovery figures, or any other quantitative claim — for Verke's own coaches. The trial's peer-reviewed results are in the published paper; we don't restate them here.
• Any clinical-assessment capability. Verke's coaches do not screen for, label, or evaluate clinical conditions.

If you're a clinician, researcher, or journalist looking into Verke's methodology or the Stockholm study, the public sources are:

• Stockholm University research catalogue entry — project page with study design and timeline.
• carlbring.se/publikationer — Professor Carlbring's full peer-reviewed publication list, including his AI-in-psychotherapy work.
• carlbring.se/uppsatser — supervised master's theses listing, where the Hassbrink quantitative thesis is catalogued.
• Stockholm University via DiVA-portal.org — the Swedish academic-thesis archive (search by author or supervisor name).
• Hlynsson, Mechler, Lindqvist, Andersson & Carlbring (2026), Anna vs. Judith: A randomized comparison of AI-delivered psychodynamic and cognitive behavioral therapies for social anxiety disorder, Internet Interventions, 45, 100960 — the peer-reviewed journal article reporting the trial.

For collaboration, methodology questions, or a deeper conversation about Verke's coaching design, write to us at support@verke.co. Response time is typically a few working days.